Changes in Genetic Diversity from Field to Laboratory during Colonization of Anopheles darlingi Root (Diptera: Culicidae)

Abstract

The process of colonizing any arthropod species, including vector mosquitoes, necessarily involves adaptation to laboratory conditions. The adaptation and evolution of colonized mosquito populations needs consideration when such colonies are used as representative models for pathogen transmission dynamics. A recently established colony of Anopheles darlingi, the primary malaria vector in Amazonian South America, was tested for genetic diversity and bottleneck after 21 generations, using microsatellites. As expected, laboratory An. darlingi had fewer private and rare alleles (frequency < 0.05), decreased observed heterozygosity, and more common alleles (frequency > 0.50), but no significant evidence of a bottleneck, decrease in total alleles, or increase in inbreeding compared with field specimens (founder population). Low-moderate differentiation between field and laboratory populations was detected. With these findings, and the documented inherent differences between laboratory and field populations, results of pathogen transmission studies using this An. darlingi colony need to be interpreted cautiously.

Because laboratory breeding/rearing environments are generally stable, unlike the wild, pathogen transmission data garnered using colony organisms may be biased. Under laboratory conditions, genetic alleles maintained in the wild may be selected against and lost.¹ In addition, colonized vectors are not subject to the same stochastic threats (biotic and abiotic) as those in the field. As such, gene expression, sexual development rate, and other aspects of vector physiology may be altered during colonization.^2,3 For example, analysis of field Anopheles gambiae transcriptomes found increased expression of genes associated with immunity, insecticide resistance, and olfaction compared with laboratory mosquitoes, whereas laboratory mosquitoes had elevated expression of metabolism and protein synthesis genes.² Although natural pathogen–vector interactions may not be replicated using colonized vectors (healthier overall due to increased food availability and stable conditions, and/or have immature immune systems due to sterile rearing environment), their use provides important preliminary information.

Laboratory colonies, established using field-collected individual mosquitoes, generally undergo genetic drift, selection, and/or bottleneck, and, consequently, may not be representative of the original source population. Post-colonization genetic diversity can be affected by loss of rare alleles, decreasing heterozygosity and effective population size, and inbreeding, potentially affecting biological interactions with and response to a pathogen.^4–8 After 20 years in a laboratory colony, Arias and others⁹ found great genetic differentiation between laboratory and field populations of Anopheles albimanus, detected with the mitochondrial Cytb gene (F_ST = 0.37179). As time since colonization increases, genetic differences can compound. Therefore, the evaluation of the similarity between colony and field populations, and whether founder effects or bottlenecks have occurred are important before interpreting results from pathogen–vector model experiments with colonized vectors. The trade-off between the validity of the comparison between laboratory and field vectors and the resources/logistics required for the use of a natural system for these experiments needs consideration as well.⁸ Recently, our group reported the colonization of Anopheles darlingi, an important malaria vector in Amazonian South America.¹⁰ The comparison of genetic heterogeneity in colonized versus wild mosquitoes is not often studied, and doing so in previously difficult-to-colonize An. darlingi is important to investigate the possibility of a bottleneck. Here, using 14 microsatellite loci, we report that whereas the total number of private alleles, rare alleles (allelic frequency < 0.05), and observed heterozygosity have significantly decreased, and the total number of common alleles (allelic frequency > 0.50) has significantly increased over 21 generations, we did not detect a statistically significant signature of a bottleneck, decrease in total alleles, or increase in inbreeding in laboratory versus field specimens. In addition, pairwise F_ST tests showed only low-moderate differentiation between field and laboratory An. darlingi.

Adult female An. darlingi field specimens were collected using human landing catch (HLC) with two collectors in 12-hour peridomestic collections in Cahuide in 2012 (May, October) and 2013 (April, June), and identified morphologically.^11,12 Adult females from June 2013 collection (N = 135) were used to establish the laboratory colony, which is currently at the F₂₈ generation.¹⁰ The colony is fed chicken and cow blood using membrane feeders daily (generations: F₁–F₁₆) or three times weekly (generations: F₁₇–current), with egg collection 42–72 hours post-feeding. The average number of adults per generation in the colony has increased over time (F₁–F₁₀ = 1,869; F₁₁–F₂₅ = 15,396). For this study, specimens of laboratory-colonized male and female An. darlingi were obtained at generations F₅, F₁₂, and F₂₁ (N = 21, 28, and 30, respectively), over the course of 1.5 years after establishment (∼15 generations annually) for comparison with field-collected specimens from May and October 2012, and April 2013 (N = 53, 41, and 12, respectively). This study was approved by the Human Subjects Protection Program of the University of California, La Jolla, San Diego, CA, and by the Comité de Ética of the Universidad Peruana Cayetano Heredia and of Asociación Benéfica PRISMA, Lima, Peru. The New York State Institutional Review Board considers HLC to be an occupational health/risk management issue, and malaria prophylaxis was offered to collectors in accordance with this policy.

Genotyping of 14 An. darlingi-specific microsatellite loci was conducted using published primers as previously described.¹³ Genotyping of polymerase chain reaction products were carried out at the Applied Genomic Technologies Core at the Wadsworth Center, New York State Department of Health, using the ABI3730 DNA Analyzer with GeneScan™ 600 LIZ^® dye size standard (Applied Biosystems, Carlsbad, CA) and data analyzed/alleles called using GeneMapper 4.0 software (Applied Biosystems). A Microsoft Excel database of alleles was converted to compatible file formats for analysis programs, and the total number of alleles (A), private alleles (A_P), and rare (allele frequency < 0.05) and common (allele frequency ≥ 0.50) alleles per locus were calculated using CONVERT v.1.31 (West Lafayette, IN).¹⁴ Observed heterozygosity (H_O), linkage disequilibrium (LD), and measures of differentiation (F_ST) and inbreeding (F_IS) were calculated using Arlequin v.3.5 (Bern, Switzerland).¹⁵ Differences in individual genotypes were visualized with factorial correspondence analysis (FCA) in GENETIX v.4.05.2 (Montpellier, France).¹⁶ The nominal significance level (α = 0.05) was adjusted for pairwise comparisons using Bonferroni correction, and repeated measures analysis of variance (ANOVA) was used to compare measures of diversity by collection, with paired microsatellite loci.

Genetic differences between the colony and field mosquitoes were evaluated using basic measures of diversity and allele frequency (Table 1). A significant decrease in A_P was detected (P = 0.012) in colony versus field specimens, though no post hoc comparisons were significant after Bonferroni correction (Table 1). The total number of common and rare alleles differed significantly by collection, that is, May 2012 specimens had significantly fewer common alleles than F₂₁ generation (ANOVA P = 0.000821, post hoc t test P = 0.017; Table 1). The frequency of rare alleles was significantly lower with the greatest difference between F₂₁ generation and May 2012 specimen (ANOVA P = 9.23 × 10⁻⁶, post hoc t test P = 0.0047), followed by May 2012 and April 2013 specimens (P = 0.0138), possibly due to low sample size in the latter, and October 2012 specimen and F₂₁ generation (P = 0.0167; Table 1). Similar changes in microsatellite alleles have been documented previously in laboratory-colonized An. gambiae s.s.⁶ Rare alleles, relatively frequent in wild populations, are often lost after colonization, because of genetic drift, and are replaced by increasing numbers of common alleles.⁶ Significant differences were detected in neither the total number of alleles per collection (A) nor the inbreeding coefficient (F_IS) among collections (ANOVA, P = 0.0669 and P = 0.9, respectively; Table 1).

Table 1.

Number of Anopheles darlingi (N), alleles (A), private alleles (A_P), alleles with specified allelic frequencies, inbreeding coefficient (F_IS), and Wilcoxon signed rank and mode shift tests for bottlenecks using 14 microsatellite loci, by source population

	Field An. darlingi			Colony An. darlingi
	May 12	October 12	April 13	F5	F12	F21
N	53	41	12	21	28	30
An.s.	113	104	80	72	73	62
AP*	12	6	4	5	4	1
A frequency ≥ 0.50*	2†	4	3	7	7	9†
A frequency < 0.05*	48†‡	34§	19†	25	19	10‡§
FISn.s.	0.47	0.48	0.52	0.61	0.57	0.63
Wilcoxon test
TPM∥n.s.	0.0083	0.0083	0.0148	0.0676	0.1206	0.1206
SMMn.s.	0.9973	0.9877	0.8662	0.9979	0.9324	0.8794
Mode shift	Normal	Normal	Normal	Normal	Normal	Normal

ANOVA = analysis of variance; SMM = stepwise mutation model; TPM = two-phase model.

^*Statistically significant differences by collection, over all 14 loci in A_P, A frequency ≥ 0.50, and A frequency < 0.05 (repeated measures ANOVA, P = 0.012, P = 0.000821, and P = 9.23 × 10⁻⁶, respectively).

^{†, ‡, §}Statistically significant post hoc, Bonferroni-corrected pairwise t tests.

^∥TPM settings: proportion of SMM in TPM = 0; variance of geometric distribution for TPM = 36 [per Bottleneck software website (http://www1.montpellier.inra.fr/CBGP/software/Bottleneck/pub.html)].

^n.s.No statistically significant differences by collection, over all 14 loci in A, nor F_IS (ANOVA, P = 0.0669, and P = 0.9, respectively).

There was no evidence of a bottleneck in the process of laboratory colonization, as detected by the two-phase model (TPM) and stepwise mutation model (SMM) Wilcoxon signed rank one-tailed tests for heterozygosity excess or the mode shift test (Table 1). An increase in LD has previously been reported in populations that have experienced recent bottlenecks.^17,18 Each collection was tested for LD, and April 2013 had one loci pair in LD, whereas F₁₂ and F₂₁ had three and two pairs in LD, respectively, with one shared pair of loci between the colony generations. Greater differences were observed when collections were grouped by source, with three and 11 pairs of loci in LD in field and colony collections, respectively. This finding suggests that laboratory colonization resulted in some reduction of genetic variation, despite Wilcoxon signed rank test results to the contrary (Table 1). Mean H_O (± one standard error [SE]) over all loci were plotted for visual comparison by collection (Figure 1 ). A statistically significant difference in H_O was detected (P = 0.00068) among collections with the greatest difference between May 2012 specimen and both F₅ and F₂₁ generations (post hoc t tests P = 0.0245 and 0.0093, respectively). According to FCA analysis, field and laboratory specimens formed two distinct clusters, with little overlap (Figure 2 ). Within these clusters, there was an overlap of specific alleles among the collections. Moderate differentiation was detected when collections were placed into field and colony groupings (F_ST = 0.05129, P < 0.0001). F_ST pairwise comparisons among collections ranged from 0.01722 (October 2012 specimen versus April 2013 specimen) to 0.10663 (April 2013 specimen versus F₅ generation). In general, the largest pairwise F_ST values were seen in comparisons between field and colony collections, whereas comparisons within each group ranged from 0.01722 to 0.03058 (field), and from 0.03062 to 0.08986 (colony).

Figure 1.

Mean observed heterozygosity for 14 microsatellite loci of field and colony Anopheles darlingi by collection date or colony generation. A statistically significant difference was detected among collections (repeated measures analysis of variance [ANOVA], P = 0.00068) with a significant Bonferroni-corrected post hoc pairwise t tests between May 2012 specimen and both F₅ and F₂₁ generations (P = 0.0245 and 0.0093, respectively). Error bars represent one standard error (SE).

Figure 2.

Factorial correspondence analysis of field-collected and colony Anopheles darlingi. Mosquitoes grouped by source: field (gray) or colony (white). Axes 1 and 2 explain 39.17% and 22.42% of the inertia in individual An. darlingi genotypes, respectively. CAH = Cahuide, Loreto, Peru.

Our data suggest that the An. darlingi laboratory colony is in the process of differentiation from the original wild population after 21 generations, despite there being no significant evidence of a bottleneck, decrease in A, or increase in F_IS. Other researchers have suggested strategies for maintaining genetic and phenotypic diversity within laboratory colonies by exposing vectors to semi-field conditions¹⁹ or by introducing wild males.²⁰ A recent study compared the levels of genetic diversity and inbreeding in An. gambiae s.l. mosquitoes collected from the field, or from colonies in semi-field cages or insectaries within the semi-field cages.¹⁹ These findings indicated that the use of semi-field cages maintained higher levels of genetic diversity and lower levels of inbreeding, compared with the insectary, and an average body size similar to that of field populations.¹⁹ Semi-field colonies may be important and necessary for studying the natural context of pathogen–vector interactions under real-world environmental conditions.